E ﬀ ects of a Salicylic Acid Analog on Aphis gossypii and Its Predator Chrysoperla carnea on Melon Plants

: The salicylic acid analog BTH (benzo-(1,2,3)-thiadiazole-7-carbothioic-acid S-methyl ester) induces systemic acquired resistance by promoting plant resistance against numerous plant pathogens and some insect pests. The objective of the research was to evaluate the activation of plant defenses with BTH on melon ( Cucumis melo L., Cucurbitaceae) and its e ﬀ ects on the herbivore Aphis gossypii Glover, 1877 (Hemiptera: Aphididae) and on the aphid predator Chrysoperla carnea (Stephens, 1836) (Neuroptera: Chrysopidae). Under laboratory conditions, plants were sprayed with BTH (50 g / ha) zero (B0), four (B4), and seven (B7) days prior exposure to insects. B0 treatment resulted in 100% mortality of aphid nymphs and disrupted adult feeding behavior (recorded by electrical-penetration-graphs technique), by prolonging the time to reach the phloem, requiring more probes to ﬁrst salivation and reducing ingestion activities. There were no e ﬀ ects on feeding behavior of A. gossypii fed on B4 plants but on its life history because fewer nymphs were born, intrinsic rate of natural growth decreased, and mortality increased. There were no e ﬀ ects on biological parameters of aphids reared on B7 plants. Prey consumption by C. carnea larvae when predated A. gossypii fed on BTH-treated plants was not di ﬀ erent among treatments. Therefore, BTH enhances the suppression of A. gossypii in the short term, without negative e ﬀ ects on the predatory larva C. carnea , which makes this plant strengthener a useful tool to be considered in integrated pest management programs. Aphid biological parameters were studied comparing B4, B7, and control treatments for the second trial. Aphid probing and feeding behavior as well as feeding efficiency of C. carnea were studied comparing B4, B0, and control treatments. All data sets that followed the assumptions of normality and homoscedasticity were analyzed by one-way analysis of variance (ANOVA). When necessary, data were transformed using the SQRT (X + 1) to reduce heteroscedasticity. The transformed data that also followed a Gaussian distribution were analyzed by one-way ANOVA test, while the non-parametric Kruskal–Wallis test was used for variables that did not follow a normal distribution or remained heteroscedastic. Subsequently, when significant differences were found ( p < 0.05) in the ANOVA, a post-hoc Tukey test was performed for pairwise comparisons; likewise, Kruskal–Wallis test was followed by post-hoc Dunn’s pairwise comparison test with Bonferroni correction. Statistical tests were conducted using the IBM SPSS Statistics 24.0 software (package for Windows, 64-bit edition, Chicago, IL, USA) at a 0.05 significance level. A chi-squared test (if expected values were higher than 5) or a Fisher´s exact test if expected values were lower than 5 (Stat View 2, Abacus Concepts, Berkeley, CA, USA) were used to analyze the PPW among treatments.


Introduction
Plants respond to herbivores and pathogens with a wide range of physical and chemical defenses [1,2]. The last can be constitutive, inducible, and even indirect through the use of nectars and volatile organic compounds (VOCs) that attract natural enemies of pests [3,4]. Chemical defenses involve complex processes with cascading reactions that use molecules, signals, and regulators with an initial short-term response at membrane level when the plant receptors detect the presence of phytophagous pests and the subsequent trigger of induced direct and indirect defenses [4]. Systemic acquired resistance (SAR) is a type of induced long-lasting defense response [5], very effective against biotrophic and hemibiotrophic pathogens and some insect herbivores [6]. The induction of SAR is signaled by the phytohormone salicylic acid (SA) and it is associated with pathogenesis-related proteins Melon cv. Sancho (Syngenta Seeds B.V., Enkhuizen, The Netherlands), a hybrid melon cultivar (toad skin type) widely planted in Central Spain was used for the experiments.
The clonal A. gossypii colony was originally started from a single virginoparous apterae collected from melon in El Ejido, Spain, in 1998. Laboratory colony was reared on melon plants and renewed every fortnight inside plastic aphid-proof cages in walk-in environmental chambers (23:18 • C temperature, 75% relative humidity (RH), and 16 Chrysoperla carnea larvae were initially purchased from Agrobio (La Mojonera, Almería, Spain) and maintained for several generations in a climatic chamber (25 ± 2 • C, 75 ± 5% RH and 16:8 (L:D) photoperiod) at the ETSIAAB-UPM laboratory, following standard rearing procedures [31]. To obtain third larval stage L3 (<24 h) for the experiments, a gauze was put on top of the adult rearing plastic cage to allow mated females to lay eggs for 24 h. Then, the gauze with the eggs was placed in a new cage with zig-zag filter paper and Ephestia kuehniella (Zeller) eggs ad libitum (as food for newly-emerged larvae). Once larvae reached L2 stage, they were individually placed with E. kuehniella eggs ad libitum to avoid cannibalism. Change in the larval stage was verified with the molt presence.
Aphids were reared in a controlled environmental chamber (SANYO-MLR 351, Osaka, Japan) at 23-19 • C, 75% RH, and 16:8 h (L:D) photoperiod and synchronized for the experiment. Then, to guarantee same-aged aphids, adult females from the laboratory colony were placed on 5-week-old melon plants. Aphids were maintained for 24 h inside clip-cages (3 cm diameter) with a mesh at the top to allow ventilation, then adults were removed with a wet paintbrush and the new nymphs were left on plants to develop into adults until the bioassay.
To analyze the effect of BTH on the biological parameters of A. gossypii, a single adult apterae female (8-9 days old since birth) was placed on the adaxial side of the second true-leaf of 5-week-old melon plants. This day was considered the beginning of the bioassay. Aphid females were confined for 24 h to generate offspring in a 3 cm diameter clip-cage. Afterward, only one neonate nymph was left per plant, which was monitored until adulthood. Then, adult offspring were counted by removing its nymphs daily until adults complete a period equal to the pre-reproductive period (number of days from birth to the onset of its reproduction) [33,34]. Aphid survivorship, duration of each nymphal instar (presence of exuvia was indicative of molting), pre-reproductive period (d), effective fecundity (offspring produced during period equal to pre-reproductive period) (Md), intrinsic rate of natural increase (rm = 0.738 (ln Md)/d) [35,36], mean generation time (Td = d/0.738), and mean relative growth rate (RGR = rm/0.86) [37] were calculated. Plants' position inside the environmental chamber was randomized every day to avoid any positional effect. Each replicate consisted of a single aphid and plant. Due to the limited space inside the environmental chamber, in the first experiment, control, and B0 treatments were compared (n = 34 replicates, respectively), and in a second one, control, B4 and B7 treatments (n = 22, 25, and 25 respectively). For the B0 treatment, only aphid survival was analyzed as all nymphs died before the sixth day of monitoring.

Aphis gossypii Feeding Behavior
Five-week-old melon plants were used in this experiment conducted at ICA-CSIC. Melon was seeded every week to have enough plants for the trials. Seven-day-old seedlings were transplanted into 1:1 mixture of soil substrate (GoV4, Jiffy International, A.S., Stange, Norway) and vermiculite Aphis gossypii colonies were reared on melon plants inside rearing cages placed in an environmental chamber at 23:18 • C (L:D), 75% HR, and 16:8 h (L:D) photoperiod. Aphids were synchronized to guarantee age homogeneity at the time of the electrical penetration graph (EPG) recording.
To determine whether the application of BTH to melon plants affected A. gossypii feeding behavior, the EPG technique [38] was used to monitor in real-time the plant penetration activities of the aphid pierce-sucking mouthparts, adapting the method previously described for A. gossypii on cucumber (Cucumis sativus L.) [33]. Plants from the three treatments (control, B0, and B4), treated as described above, were used every time EPGs were performed. Because there were no statistical differences between aphid fitness on control plants and on B7, this treatment was not included in the feeding behavior experiment.
To facilitate wiring, apterous adult A. gossypii (11-12 days old since birth) was immobilized using a vacuum-operated plate (Eyela Aspirator A3S, Tokyo Rikakikai Co. Ltd., Tokyo, Japan). Then, an extra thin gold wire (2 cm length, 18.5 µm in diameter) was attached to the aphid dorsum with a small droplet of water-based silver-conducting glue (EPG-Systems, Wageningen, The Netherlands). The opposite extreme of the gold wire was glued to a thin copper wire (2-cm length), attached to a brass pin, which was inserted into the input connector of the primary amplifier of the 4-channel Gigaohm DC-EPG device (EPG Systems, Wageningen, The Netherlands). Another copper electrode (10-cm length, 2-mm diameter) was inserted into the soil of the plant container. Aphids were placed on the abaxial side of the third of four-leaf growth-stage melon plants. Each single aphid and melon plant was used only once for EPG recordings. The probing and feeding behavior was recorded during a 6-h period, which started directly after aphids were placed on the melon leaves. The EPGs acquisition was carried out inside a Faraday cage to avoid electrical noises in an air-conditioned room (24 • C). Data acquisition and analysis were conducted by Stylet + v01.25 software for Windows (EPG Systems, Wageningen, The Netherlands). Finally, 20 replicates per control treatment and 24 replicates per B0 and B4 were analyzed.
The aphid associated EPG waveforms [38] analyzed in this study were: (non-probe) non-probing behavior, i.e., no stylet contact with the leaf tissue; (C) intercellular apoplastic stylet pathway where insects show a cyclic activity of mechanical stylet penetration and secretion of saliva; (E1) salivation into phloem sieve elements at the beginning of the phloem phase; and (E2) passive phloem sap uptake from the sieve elements. The term "probe" refers to any type of event during the period in which the stylets of an individual insect were in contact with plant tissue, whereas "non-probe" refers to the event with no contact between stylets and plant tissue [39]. EPG sequential and non-sequential variables related to the probing and feeding behavior, described by the different waveforms, were calculated for each EPG recording using the MS Excel workbook for automatic parameter calculation of EPG data (version 5.0) [40]. The mean ± standard error of the 32 selected EPG sequential and non-sequential variables was calculated and organized as described in Backus et al. [41]: PPW, the proportion of individuals that produced a specific waveform type; NWEI, number of waveform events per insect, that is the sum of the number of events of a particular waveform divided by the total number of insects under each treatment; WDI, waveform duration per insect, calculated using the sum of the duration of each event of a particular waveform made by each individual insect that produced the waveform divided by the total number of insects under each treatment; and WDE, waveform duration (min) per event, that is the sum of the duration of the events for a particular waveform divided by the total number of events of that particular waveform under each treatment (this parameter provides information on behavior at the population level).

Feeding Efficiency of Chrysoperla carnea
A protocol adapted from Wanumen et al. [42] was used to analyze whether BTH had a repellent or anti-feeding action on the predator C. carnea when the prey A. gossypii fed on treated plants, as some alterations in aphid feeding behavior on BTH-treated plants had been observed previously. The L3 C. carnea was fed for three consecutive days with four instar A. gossypii nymphs (N4), previously fed over 6 h on BTH-treated plants, as changes in aphid feeding behavior had been already observed over this period in the prior EPG experiment. This instar was used as prey for C. carnea instead of adults, in order to know the exact number of prey ingested/day, because the latter could generate offspring. Accordingly, the aphid laboratory colony reared in ETSIAAB-UPM facilities was synchronized to have enough 5-day-old nymphs (N4) for the bioassay.
Melon seedlings were transplanted in 9 × 9 × 10 cm pots and maintained in the growth chamber at ICA-CSIC facilities as described in A. gossypii feeding behavior experiment until bioassay in ETSIAAB-UPM laboratory when plants were five weeks old. Thereafter, plants were maintained in the environmental chamber (SANYO-MLR 351) at the same environmental conditions as in ICA-CSIC. Plants from the three treatments (control, B0, and B4) were used in the bioassay. Three plots of plants per treatment, one for each day of consumption experiment, were sprayed at the same time (21 plants/treatment). The fourth true-leaf of each plant was placed in 12 cm cylindrical cages, with a hole with foam between the lid and the cage to protect and allow the petiole insertion. In this way, BTH could continue relocating through the plant as cutting the leaf was avoided. Filter paper was placed at the bottom of the cage to absorb moisture and prevent water condensation.
Three consecutive aphid synchronizations were performed to get enough nymphs for the experiment (300 adults to get ≈ 2400 N4/day, to use ≈ 1400). Daily, 21 lots of 60 N4 aphids (the two first days) or 65 (the third day), as L3 were bigger and could need to ingest more food supply, were placed in 4 cm cages with ventilation (7 replicates × 3 treatments), where they were fasting until passed to the treated plants. Aphids were transferred to the 12 cm cage and allowed to settle on the leaf. Nymphs were feeding on the plant for 6 h. After this period, one L3 C. carnea (<24 h), that had been starved for 24 h to assure it had a voracious appetite, was placed on the cage and left 24 h to predate the nymphs. After 24 and 48 h later, the same procedure was repeated with new plants and aphids but transferring the same C. carnea larvae.
Every 24 h, once larvae had been transferred to new experimental units, leaves with aphids (predated or not) were excised and aphids counted. The fourth day, C. carnea larvae were placed in 2.5 cm diameter cages and fed with untreated E. kuehniella eggs ad libitum until larvae pupated. Larvae were checked daily until pupation and adult emergence. Furthermore, C. carnea larvae were weighed immediately before the beginning of the bioassay (0 h) and 72 h after feeding on aphid nymphs.
The parameters evaluated were: L3 C. carnea larval development time until pupation (days), pupation length (days), daily (after 24, 48, and 72 h) and total consumption of aphids (number of aphids ingested/offered), and total increase in weight after three days of contaminated prey consumption (mg). The experiment was repeated three times.

Statistical Analysis
Aphid survivorship data were analyzed by Kaplan-Meier survival analysis with global and pairwise multiple comparison procedures in order to compare survival curves among treatments Agronomy 2020, 10, 1830 6 of 17 (Log-Rank test, p < 0.05). Censored observations were those subjects that had not died at the end of the study and those that disappeared midway in the study (aphids that escaped from clip-cages) [43].
Aphid biological parameters were studied comparing B4, B7, and control treatments for the second trial. Aphid probing and feeding behavior as well as feeding efficiency of C. carnea were studied comparing B4, B0, and control treatments. All data sets that followed the assumptions of normality and homoscedasticity were analyzed by one-way analysis of variance (ANOVA). When necessary, data were transformed using the SQRT (X + 1) to reduce heteroscedasticity. The transformed data that also followed a Gaussian distribution were analyzed by one-way ANOVA test, while the non-parametric Kruskal-Wallis test was used for variables that did not follow a normal distribution or remained heteroscedastic. Subsequently, when significant differences were found (p < 0.05) in the ANOVA, a post-hoc Tukey test was performed for pairwise comparisons; likewise, Kruskal-Wallis test was followed by post-hoc Dunn's pairwise comparison test with Bonferroni correction. Statistical tests were conducted using the IBM SPSS Statistics 24.0 software (package for Windows, 64-bit edition, Chicago, IL, USA) at a 0.05 significance level.
A chi-squared test (if expected values were higher than 5) or a Fisher's exact test if expected values were lower than 5 (Stat View 2, Abacus Concepts, Berkeley, CA, USA) were used to analyze the PPW among treatments.

Aphis gossypii Individual Fitness
Survival curves of A. gossypii varied significantly among treatments (Log rank test, global comparisons: χ 2 = 112.643, df = 4, p ≤ 0.001) and that of B0 plants (BTH fresh residue) significantly differed from the rest. This treatment caused 100% mortality of A. gossypii nymphs in the first six days of evaluation and the individual fitness parameters could not be analyzed. The survival curve of A. gossypii reared on B4 plants (treated with BTH four days before the bioassay when the initial aphids were placed on treated plants) was significantly different from that of B7 plants. Survivorship of aphids on B4 and B7 plants was similar to that of the control of its trial (C2) (Log rank test, pairwise multiple comparison; Figure 1, Supplementary Table S1). Although no significantly different, C2 survival curve was lower than B7, this could be explained as there were some censored observations in C2 treatment, as some aphids disappeared from clip-cages.
Agronomy 2020, 10, x FOR PEER REVIEW 6 of 17 Aphid biological parameters were studied comparing B4, B7, and control treatments for the second trial. Aphid probing and feeding behavior as well as feeding efficiency of C. carnea were studied comparing B4, B0, and control treatments. All data sets that followed the assumptions of normality and homoscedasticity were analyzed by one-way analysis of variance (ANOVA). When necessary, data were transformed using the SQRT (X + 1) to reduce heteroscedasticity. The transformed data that also followed a Gaussian distribution were analyzed by one-way ANOVA test, while the non-parametric Kruskal-Wallis test was used for variables that did not follow a normal distribution or remained heteroscedastic. Subsequently, when significant differences were found (p < 0.05) in the ANOVA, a post-hoc Tukey test was performed for pairwise comparisons; likewise, Kruskal-Wallis test was followed by post-hoc Dunn's pairwise comparison test with Bonferroni correction. Statistical tests were conducted using the IBM SPSS Statistics 24.0 software (package for Windows, 64-bit edition, Chicago, IL, USA) at a 0.05 significance level.
A chi-squared test (if expected values were higher than 5) or a Fisher´s exact test if expected values were lower than 5 (Stat View 2, Abacus Concepts, Berkeley, CA, USA) were used to analyze the PPW among treatments.

Aphis gossypii Individual Fitness
Survival curves of A. gossypii varied significantly among treatments (Log rank test, global comparisons: χ 2 = 112.643, df = 4, p ≤ 0.001) and that of B0 plants (BTH fresh residue) significantly differed from the rest. This treatment caused 100% mortality of A. gossypii nymphs in the first six days of evaluation and the individual fitness parameters could not be analyzed. The survival curve of A. gossypii reared on B4 plants (treated with BTH four days before the bioassay when the initial aphids were placed on treated plants) was significantly different from that of B7 plants. Survivorship of aphids on B4 and B7 plants was similar to that of the control of its trial (C2) (Log rank test, pairwise multiple comparison; Figure 1, Supplementary Table S1). Although no significantly different, C2 survival curve was lower than B7, this could be explained as there were some censored observations in C2 treatment, as some aphids disappeared from clip-cages.  Survival curves of Aphis gossypii reared on melon plants sprayed with BTH (benzo-(1,2,3)-thiadiazole-7-carbothioic-acid S-methyl ester) to the point of run-off, zero (B0), four (B4), and seven (B7) days prior to aphid introduction (Log rank test, p-value < 0.05). Control plants were sprayed with distilled water (Control 1: first trial control, Control 2: second trial control) the day of aphid introduction.
The development time of each nymphal instar, the pre-reproductive period as well as the mean generation time of A. gossypii were not affected by the application of BTH in B4 and B7 plants (p > 0.05 in all cases, Table 1). The effective fecundity (Md) of aphids grown on B4 plants significantly decreased (23.16%) compared to the control, but the effect disappeared when aphids were reared on B7 plants (Md: H = 7.565, p = 0.023) ( Table 1).
There was a significant reduction in the intrinsic rate of natural increase (rm) of aphids grown on B4 plants compared to the control and to aphids grown on B7 plants. There were no differences in rm between aphids reared on B7 plants and the control. Because of the mean relative growth rate (RGR) is a related to rm by the formula (rm = 0.86 RGR), the same trend was observed (rm and RGR: F 2,42 = 8.371, p = 0.001) ( Table 1).

Aphis gossypii Feeding Behavior
The effect of BTH on the feeding behavior of aphids fed on B0 and B4 plants was analyzed because mortality of aphids exposed to the first treatment was very high and fecundity and growth rates were affected on B4 plants. However, as there were no differences in the biological parameters of A. gossypii reared on B7 plants compared to the control, this treatment was not considered.
In general, for the majority of variables studied, feeding behavior of aphids exposed to control plants (C) did not differ significantly from the other two treatments (B0-B4) ( Table 2). The probing activities measured by the number of waveform events per insect (NWEI) of "probe", intercellular stylet pathway "C" and "short probes" (C < 3 min), were significantly more numerous in aphids feeding in B0 plants compared to B4 ones (p < 0.05 in all cases, Table 2). The duration of the "probe" waveform per insect (WDI) was significantly shorter while the duration of "C" was significantly longer on aphids feeding in B0 plants compared to B4 ones (p < 0.05 in both cases, Table 2). No differences were observed in the duration of the "probe" event as well as in the "C" event (WDE) among treatments (WDE, p > 0.05 in both cases, Table 2). The percentage of probing time spent in intercellular stylet pathway activities ("C") was significantly higher in aphids feeding in B0 plants compared to those feeding in B4 ones (p < 0.05, Figure 2). There were significantly more "non-probe" events (NWEI) and its durations per aphid (WDI) and per event (WDE) were significantly longer in aphids feeding in B0 plants compared to those feeding in B4 plants (p < 0.05 in all cases, Table 2).
Sequential EPG variables that describe the sequence of events related to each other during the six hours of recording are shown in Table 3. Time spent from first probe to first E, time from first probe to first sE2, time from first probe to first E2 and number of probes to the first E1 were significantly higher in aphids feeding in B0 plants than in the other two treatments, not observing differences in these variables between aphids feeding on B4 and control plants (p < 0.05 in all cases, Table 3). Total duration of non-probe before the first E was significantly longer in aphids feeding in B0 plants than in B4 plants, while aphids exposed to control plants did not significantly differ from the other two treatments (p < 0.05, Table 3). Time to first probe from start of EPG, time from the beginning of that probe to first E, time from the beginning of that probe to first E2 and time from the beginning of that probe to first sE2 were not significantly different among treatments (p > 0.05 in all cases, Table 3).  -(1,2,3)-thiadiazole-7-carbothioic-acid S-methyl ester) to the point of run-off, zero (B0) and four (B4) days prior to aphid first contact, using water-sprayed plants as control. Six-hour electrical penetration graph (EPG) recording. Waveforms: C, intercellular stylet pathway; E shows phloem-related activities: E1, correlates with salivation into phloem sieve elements; E2, regards as phloem ingestion [44]. Statistical comparisons among treatments for each parameter were made by non-parametric Kruskal-Wallis test. Means within variables followed by different lower-case letters are significantly different (p-value < 0.05).
Sequential EPG variables that describe the sequence of events related to each other during the six hours of recording are shown in Table 3. Time spent from first probe to first E, time from first probe to first E2, time from first probe to first E2 and number of probes to the first E1 were significantly higher in aphids feeding in B0 plants than in the other two treatments, not observing differences in these variables between aphids feeding on B4 and control plants (p < 0.05 in all cases, Table 3). Total duration of non-probe before the first E was significantly longer in aphids feeding in B0 plants than in B4 plants, while aphids exposed to control plants did not significantly differ from the other two treatments (p < 0.05, Table 3). Time to first probe from start of EPG, time from the beginning of that probe to first E, time from the beginning of that probe to first E2 and time from the beginning of that probe to first sE2 were not significantly different among treatments (p > 0.05 in all cases, Table 3). Table 3. Mean ± standard error (min) of sequential electrical penetration graph (EPG) variables and their ranges in parenthesis for the feeding behavior of Aphis gossypii exposed to melon plants at 0 (fresh residue) (B0) and 4 days (B4) after sprayed BTH (benzo- (1,2,3)-thiadiazole-7-carbothioic-acid S-methyl ester), using water-sprayed plants as control, during a six-hour recording.

Discussion
Aphis gossypii has evolved resistance to the principal families of synthetic insecticides [27,45]. Owing to the fact that the application of pesticides must be reduced to achieve sustainable agriculture, the use of elicitors as plant strengtheners that could induce herbivore resistance at the same time that enhance the biological control of these pests could be an interesting strategy to be included in IPM. The BTH promotes pathogen resistance in a wide range of crops [6,14], but just a few studies have explored the susceptibility of herbivore insects and mites to this elicitor [15][16][17]46]. It is known that BTH improves indirectly the biological control since it can increase the attractiveness of herbivore-damaged plants to parasitic wasps due to qualitative and quantitative changes in the induced volatiles emissions [19][20][21], but information on BTH-effect on predators is insufficient.
There is great heterogeneity in scientific literature about how and when is necessary to apply BTH to induce plant resistance, and the time and mode of application chosen influence plants and herbivorous insects differently. BTH might be applied soaking seeds [47], spraying leaves or the whole plant [20], or as a soil drench [21]. Some researchers apply the product in young plants (1-2 weeks old) while others in older plants and just a few hours/days before the starting of the bioassays [16,18,46,48]. In this study, the pest was exposed to three different application periods of BTH to discern if the effect on insects' performance was due to an induction of defenses or due to a direct toxicity of the product.
A reduction in foliar thickness that finally generated necrotic lesions on BTH-sprayed leaves was observed in this experiment in agreement with phytotoxic effects detected in different plant species after application of elicitors [16,49]. Even if SA analogs are less phytotoxic than the application of SA [6,7], BTH may cause an allocation fitness cost [5]. This compound sprayed on seedlings of melon var. Orange Flesh caused an inhibition in growth from 6 days after treatment onwards and this could be related to the peroxidase activity increase. This defense-related enzyme is associated with the lignification of the plant cell wall that could increase resistance to pathogen attack, therefore, even if BTH could reduce melon growth, it finally induces its plant defenses [24]. Further biochemical research could elucidate if the BTH-effects observed in our trophic system may be related to the possible phytotoxicity.
In the present work, the EPG technique reveals that there was not a contact effect of BTH on the initial probing behavior (time from the start of the EPG recording until the first probe) of A. gossypii exposed to the different BTH-treated plants compared to the control plants. This EPG variable is associated with pre-phloem resistant factors present in epidermis or mesophyll that reduce attractiveness of the plant to the insect [50] and is used to evaluate the deterrent effect or delays in aphid probing on insecticide-treated plants [51]. In our case, BTH neither inhibited nor delayed A. gossypii initial probe on treated plants.
In agreement with Cooper et al. [17], who reported that BTH could reduce the population growth of Macrosiphum euphorbiae Thomas, 1878 on a susceptible tomato cultivar and enhance the aphid resistance in a resistant cultivar, the present research showed that fresh residues of BTH (B0 plants) prevented the population growth of A. gossypii in melon cv. Sancho because nymphs did not develop to maturity. The B0 plants modified probing and feeding behavior of A. gossypii adults thus, aphids spent more time in non-probe activities and although they could reach the phloem, they took longer to contact the sieve elements, needed more probes to salivate for the first time and finally spent less time in ingestion activities (i.e., passive phloem sap uptake). Therefore, B0 plants could have activated their defense mechanisms at phloem level in a short term, as A. gossypii cannot initiate ingestion from the phloem as when fed on control plants or B4 plants [52]. Using the same EPG recording technique, Powell and Hodge [53] observed that phloem sap ingestion was also reduced when Acyrthosiphon pisum Harris, 1776 fed on tic bean (Vicia faba var. minor L.) treated with the elicitor β-aminobutyric acid (BABA), but only when there has been a previous aphid infestation on the plants. In our study, the reduction in A. gossypii individual fitness in addition to the impairing in its feeding behavior on B0 plants could be more related to a direct toxic effect of fresh residue of BTH or to the increase in levels of pathogenesis-related proteins gene expression (immediate-early gene induction) [11,14,54].
In B4 plants, there was no effect on A. gossypii feeding behavior compared to the control. However, there was a clear effect on aphid life history traits because fewer nymphs were born, the intrinsic rate of natural growth decreased (also, the RGR), and the mortality increased compared to those on control plants. Therefore, an antibiosis more than an antixenosis mechanism could have reduced A. gossypii population growth when fed on these plants. Despite using different methodologies, similar effects on aphid fitness were reported in other elicitor-plant-aphid systems [16,55].
BTH can induce a long-lasting resistance against some aphids [7,20], but this was not the case in the present study because A. gossypii had similar performance and survival when reared on B7 plants compared to the control. Overall, the BTH effects observed could be due to an activation of the defense mechanism only in the short term. For reasons still unknown, BTH does not induce prolonged resistance mechanisms against A. gossypii. On the other hand, BTH might be toxic to these aphids.
In this study, spray application of BTH to melon plants did not affect the feeding efficiency of C. carnea when predated A. gossypii fed on treated plants, independently of the application day. Similar to the results exposed here, the elicitor methyl jasmonate when applied to tomato seeds did not alter the behavior of Chrysoperla externa (Hagen, 1861) although reduced the performance of its pest, the leaf miner Tuta absoluta (Meyrick, 1917) [56]. In tritrophic systems involving elicitors and parasitoids from genus Aphidius, slight effects on the wasps' emergence were observed depending on the concentration of the compound, so specific studies case by case should be performed [49,57].
Alterations in aphid feeding behavior on BTH-treated plants could also change acquisition, retention, latency, and inoculation of aphid-borne plant viruses. The greater number of short superficial probes and intracellular stylet punctures increased the transmission of a non-persistent virus [33]. On the other hand, circulative virus particles are usually ingested from phloem sieve elements [58]. Therefore, stylets penetration behavior could help to determine the effect of elicitors on the transmission efficiency of plant viruses by piercing-sucking insects [59].

Conclusions
The effects of BTH applied at fresh residue could be due to a direct toxicity of the product on the aphid that undermines its fitness and feeding behavior, causing total mortality of A. gossypii nymphs. When the product is applied four days before A. gossypii exposure, BTH impairs aphid fitness reducing the population growth. A short-term but not long-lasting activation of plant defenses could be pointed for seven days after BTH application the product seems to have degraded regarding the effect on the aphid. The direct effect of BTH to A. gossypii in the short term, together with the lack of effect of the elicitor on C. carnea feeding efficiency, are indicative that this plant strengthener contributes to enhance the suppression of the phloem-feeding insect in the trophic system studied without harming the beneficial chrysopid. However, in order to consider BTH as an interesting strategy to control aphid pests within IPM for melon crops, further research should be addressed to ascertain the molecular and biochemical mechanisms behind the observed effects on melon and aphids caused by the elicitor BTH. Thus, a more efficient use of the product could be implemented.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4395/10/11/1830/s1, Table S1: Kaplan-Meier survival analysis of Aphis gossypii reared on melon plants sprayed with BTH to the point of run-off, zero (B0), four (B4), and seven (B7) days prior to aphid introduction. Control plants were sprayed with distilled water (C1: first trial control, compared to B0; C2: second trial control, compared to B4 and B7) the day of aphid introduction.